Research on core loss prediction of low-frequency transformer based on Grey Wolf optimisation algorithm optimised Back Propagation neural network

Abstract

In this paper, a prediction model of 20 Hz low-frequency transformer core loss based on the grey wolf optimisation algorithm-optimised back propagation neural network is proposed. Firstly, the loss characteristics of silicon steel sheet materials at different low-frequency temperatures and normal temperatures at different frequencies were compared. The general law of the variation of no-load iron loss with frequency and temperature is analysed. Finally, the BP neural network prediction model of low-frequency transformer core loss is established. The loss data obtained by experiment and simulation are used as training and verification samples to predict transformer core loss. The results show that the GWO-BP neural network loss model proposed in this paper successfully predicted the no-load loss of the transformer at different temperatures. When the prediction effect of the GWO-BP model was optimal, the determination coefficient R² reached 0.9169, and the mean relative error and root mean square error were only 1.15% and 0.0085, respectively. Moreover, the MRE of the GWO-BP model is within 9%. Compared with the BP model and whale optimization algorithm-BP model, the prediction accuracy of the loss is improved by the GWO-BP model, and the calculation time of the loss is reduced by the finite element method.